Any benchmarks where ReadyBoost has really boosted?

[ap2]

I'm running Vista Business on a 2.4 GHz AMD LapTop machine, 2GB of RAM.
Many of the operations are quite sticky when compared to the old XP
machine with 1 GB memory.

I was wondering if ReadyBoost would bring some speed-up to Vista. I have
seen no speed increase with my current 2 GB USB Boost sticks. Then
someone pointed me to this Tom's Hardware page there they have evaluated
Ready boost.

To me it looks like ReadyBoost would never give even to these guys more
than 2..3 percent of speed increase. And in some cases ReadyBoost even
made machines with low 1 GB memory slower than without ReadyBoost.
Any comments on this Toms's Hardware test, do they runs these tests
wrong? I did not find any Microsoft based ReadyBoost benchmarks or
anything.

Currently I am more of thinking to downgrade operationg system to XP.

I will not invest more on another 2GB of true RAM memory upgrade (not a
USB Boost) if it is uncertain if even that would speed up this laptop.
Thanks for comments.
-ap

 

[Colin Barnhorst]

Increasing ram gives a much better result since paging is reduced. Having
said that, readyboost is active on my system with 8GB of ram and there is
slight performance improvement.

 

[JP Levasseur]

On my laptop TL-56 1.8Ghz, and 2 GB Ram I was running Vista Ultimate 32 and
it was ok but not great, so I swicht to Vista Ultimate 64 it is day and
night my system is more responsive.So if you are able to make the change
for business 64 you will see a big difference.Bye JP

 

[ap2]

Increasing ram gives a much better result since paging is reduced.

I notice I had pasted wrong link in my previous message. This is the
right link to Tom's hardware and ReadyBoost tests.
http://www.tomshardware.com/2007/01/31/windows-vista-superfetch-and-readyboostanalyzed/

Having said that, readyboost is active on my system with 8GB of ram
and there is slight performance improvement.

The basic idea with SuperFetch and ReadyBoost is said to be this:

"SuperFetch analyzes your behavior and proactively puts
applications into available main memory, so they can
be launched quicker.
ReadyBoost allows expanding the main memory size by
plugging in a USB 2.0 Flash drive. "

Compared to RAM memory, USB-stick is about 20 times slower with seek
operations and maybe 50 times slower with brute transfer speeds. So
seeing boosting effect with ReadyBoost even with 8 GB of installed RAM
sounds almost like a no-so-true observation. Sorry about my
mis-believe

I mean, when you have 8 GB of ram that amount is surely enough to store
every application and all the data you have used during the last 3..5
days. Every Office Document, all the IE browsing cache, photos, MP3
music, everything.

A few memory cycles away you have every bit of apps and data you have
accessed in last 3 days. Everything readily accessible.

Yet Vista's memory management logic says that: "No, this is not an
optimal and fastest way. I'll move a few gigabytes of that RAM data to
the USB memory stick to get some boosting". That way Vista will get 2 GB
of free RAM memory so it can load and store there _even more important
data_ than that that was transferred to USB-stick ( so what data is
that?).

Even if you had 20 GB of RAM memory or even 100 GB, then Vista would
still work with that logic. It would still store some data to USB-memory
so it would "boost" your computer.

My scepticism about ReadyBoost arose at the very first time I read about
it. As a programmer (AKA) I thought that this can not be true, caching
data this way to slowish USB memory can bring only _very limited_
advantage if any.

Yet I am willing to change my opinion any day. If I someone could point
me to any Benchmark there ReadyBoost has got better results than it got
at Tom's tests.

Someone suggested to install Vista Ultimate 64 to get my slowish 2.4 GHz
Vista laptop faster. Currently my own thoughts are downgrading to XP,
stay there and wait for some years to go.

Thanks for all the comments this far.
-ap

 

[ap2]

So if you are able to make the change for business 64 you will
see a big difference.

I might be able to do that But really, voluntarily choose missing 64
driver problem, incompatibilities with current applications and
peripheral devices etc.

Currently I am not looking for the newest, partly experimental operating
systems so I could start spending time testing with them. In computer
shop I had no choice to choose for instance XP, Vista just came along as
a mandatory part.

I just wanted to find some way how to simply get over this slowish
laptop problem that frustrates me more and more every day.
Thanks anyway.
-ap

 

[Colin Barnhorst]

But it is still faster than hard drive access. In anycase, superfetch does
not depend on readyboost although they can work together.

 

[ap2]

But it is still faster than hard drive access.

Yes but ReadyBoosting from slow USB-memory is always 20 times slower
than getting the data from your 8 GB RAM memory. And that amount of RAM
can easily hold all data you have used or browsed during the last 3..4
working days.

Why on earth unload data from fast RAM and put it on a slow USB stick
for ReadyBoosting?

In anycase, superfetch does not depend on readyboost although they
can work together.

I do not even understand the difference between Vista's SuperFetch and
the standard Win32 caching and hard disk buffering system. That has been
around at least 10 years, and it has always tried to optimize and keep
cached in RAM memory all the most used data.

If ReadyBoosting would be so great technique and idea, then Vista memory
management should always separate about 15..20% of the RAM memory (about
1.6 GB of RAM for instance from your 8 GB machine) and dedicate it for
Boosting purposes only.

Boosting from fast RAM memory is always 20 times faster than boosting
from slowish USB-sticks. I do not understand even the basic logic or
maths or idea behind this boosting thing.
-ap

 

[DevilsPGD]

I do not even understand the difference between Vista's SuperFetch and
the standard Win32 caching and hard disk buffering system. That has been
around at least 10 years, and it has always tried to optimize and keep
cached in RAM memory all the most used data.

Standard disk caching is just that, a cache of what was *previously*
read (or written)

SuperFetch fetches data from disk *before* you need it, preemptively
building a disk cache.

The concept is fairly straight forward, SuperFetch will "notice" that
each morning when you log in, you start your mail client and web browser
almost immediately, so rather then waiting until you launch those
applications to read them into memory, the executable and possibly some
data files are cached in memory.

Note that the applications are not launched, this is pure disk caching.

When SuperFetch guesses right, a substantial performance increase
results, when SuperFetch guesses wrong there isn't much of a downside
since the caching process and memory allocation both back off as demand
necessitates.

If ReadyBoosting would be so great technique and idea, then Vista memory
management should always separate about 15..20% of the RAM memory (about
1.6 GB of RAM for instance from your 8 GB machine) and dedicate it for
Boosting purposes only.

Caching works very well putting otherwise unused RAM to work, but will
almost always cause a performance hit if it takes up RAM that an
application actually needs.

Locking executables or data in memory that is not being used right now
while an application actually needs memory is counterproductive, and
would result in a huge performance hit while Vista pages.

Boosting from fast RAM memory is always 20 times faster than boosting
from slowish USB-sticks. I do not understand even the basic logic or
maths or idea behind this boosting thing.

ReadyBoost's concept is to fill in the gap when you don't have enough
RAM to hold all the data that SuperFetch would like to cache, providing
some benefit over raw disk access.

 

[ap2]

SuperFetch fetches data from disk *before* you need it, preemptively
building a disk cache.

Yes, but when a user opens his/her computer in the morning the simple
90% right quess for system to make is that the user will need the same
apps and data he used during the last 2..3 days.

SuperFetch will "notice" that
each morning when you log in, you start your mail client and web browser
almost immediately, so rather then waiting until you launch those
applications to read them into memory,

At the computer startup, digging the yesterday's disk cache and the apps
and data there should be very good starting point for further quesses.

And if the RAM is as big as 8 GB, then the last 2..3 days *all data and
apps* will easily fit to the 20% portion (1.6 GB) of it.
Vista itself can use the remaining 6.4 GB for all the paging needs etc.
That should be enough for the operating system itself, as Vista will run
even on a 2 GB machine.

Locking executables or data in memory that is not being used right now
while an application actually needs memory is counterproductive, and
would result in a huge performance hit while Vista pages.

Locking or optimizing... If a Vista system has 6.4 GB of RAM memory to
use for all of the page caching and what ever it needs to do. That RAM
amount is so high that the system performance should not choke for low
memory reasons. Most of the new Vista machines for businessa use are
sold with 2 GB (even 1 GB) of RAM memory.

ReadyBoost's concept is to fill in the gap when you don't have enough
RAM to hold all the data that SuperFetch would like to cache, providing
some benefit over raw disk access.

You have never "enough" RAM if you require that the RAM size should be
as bige as the hard disk size, or even bigger.

If you have so much RAM memory that you can always keep 2..3 last days
used apps and data in a small portion (1.6 GB) of RAM memory, then you
should have good chances you can get the system swap apps and run fast
and smoothly.

I still think that it is not at all best and most optimized Vista memory
management if you with a 8 GB system have to push some *last used data*
to the slow USB stick. This kind of boosting will not boost but slow
down the data access.
-ap